What's Changed

• Dependencies were updated to address CVE-2022-21698 and CVE-2021-38561
• generate_chart.sh by @ganievs in #925

New Contributors

• @ganievs made their first contribution in #925

Full Changelog: 0.18.4...0.18.5

What's Changed

• Base image has been updated to address https://access.redhat.com/errata/RHSA-2022:0896 (closes #913)
• Https support for health checks by @bkuschel in #912
• Fixed number of SERVERS from 1 to 3 for a 3 node clickhouse-keeper deployment by @a-dot in #902
• clickhouse-keeper and ZooKeeper manifests were updated by @Slach in #911 and #916

New Contributors

• @a-dot made their first contribution in #902
• @bkuschel made their first contribution in #912

Full Changelog: 0.18.3...0.18.4

What's Changed

• Fixed a bug that might result in broken CHI if operator has been restarted during the reconcile
• Added 'AGE' to CHI status
• Added ClickHouse Keeper examples

Full Changelog: 0.18.2...0.18.3

What's Changed

• Operator now correctly respects configuration settings to exclude annotations from managed objects. Fixes #859
• Make CHI status messages more human friendly

Full Changelog: 0.18.1...0.18.2

What's Changed

• fix non pointer mistake in metrics-exporter by @adu-bricks in #870
• Helper files for operatorhub.io integration

New Contributors

• @adu-bricks made their first contribution in #870

Full Changelog: 0.18.0...0.18.1

New features

• arm64 packages are published (closes #852)
• 'access_management' can be specified when defining users in CHI.
• k8s secrets can be referenced when defining user passwords. See 05-settings-01-overview.yaml.

Changed

• When CRD is deleted, operator keeps all dependent objects now (statefulsets, volumes). That prevents from incidental delete of a cluster.
• When operator restarts it does not run a reconcile cycle anymore if CHI has not been changed. That prevents unneeded pod restarts. (Closes #855)
• Operator configuration file format has been changed. See https://github.com/Altinity/clickhouse-operator/blob/0.18.0/config/config.yaml. Configuration in old format is supported for backwards compatibility.

Fixed

• Fixed a bug 'unable to decode watch event: no kind "ClickHouseOperatorConfiguration" is registered' that could appear in some k8s configurations.
• Removed INFORMATION_SCHEMA from schema propagation. (closes #854)
• Added a containerPort to metrics-exporter (#834)

New Contributors

• @johnhummelAltinity made their first contribution in #838
• @jiangxinqi1995 made their first contribution in #846
• @johnny made their first contribution in #834

Full Changelog: 0.17.0...0.18.0

New features:

• Labels and annotations from auto templates are now supported
• Macros can be used in service annotations the same way as in generateName. Fixes #795

Changed:

• Status object has been cleaned up

Fixed:

• Database engine is now respected during schema migration
• Fixed schema migration for ClickHouse 21.11+
• Removed extra waits for single-node CHI changes
• Fixed a possible race conditions with labeling on operator startup

This is a bugfixing release with a number of internal changes:

• CRD definition of Status has been modified. Note: CRD needs to be updated with this release
• Default terminationGracePeriod for ClickHouse pod templates was moved to operator configuration. Default is 30 seconds as in 0.15.0 and before.
• Improved installation templates
• Fixed a bug with replicas not being correctly added when CHI has been modified from a Python Kubernetes client.

Upgrade notes:

• CRD needs to be updated with this release
• 0.16.0 had hardcoded 60 seconds for terminationGracePeriod that resulted in ClickHouse restarts when upgrading from 0.15.0 to 0.16.0. Upgrade from 0.15.0 to 0.16.1 does result in ClickHouse restarts. If you are upgrading from 0.16.0 to 0.16.1 set terminationGracePeriod to 60 in operator config file. Refer to Operator Configuration for more details.

New features:

• PodDisruption budget is configured automatically in order to prevent multiple pods being shut down by k8s
• topologyKey is now configurable. It addresses #772
• spec.restart: "RollingRestart" attribute has been added in order to initiate graceful restart of a cluster. Use it with a patch command.
• Added a support for Kubernetes secrets in settings section, e.g. for ClickHouse user/password.

Changed:

• Pod maintenance logic during all operations that require pod restart has been improved:
  1. Wait condition is enabled for a pod to be removed from ClickHouse cluster.
  2. Wait condition is added for running queries to complete (up to 5 minutes).
  3. terminationGracePeriod has been increased to 60 seconds.
• Timeout for DDL operations has been increased to 60 seconds. That addresses issues with schema management on slow clusters.
• Base image is switched to RedHat UBI
• ZooKeeper image version has been rolled back to 3.6.1 since we have some problems with 3.6.3
• CRD apiVersion has been upgraded from apiextensions.k8s.io/v1beta1 to apiextensions.k8s.io/v1. The v1beta1 manifests are still available under descriptive names.

Fixed:

• Fixed a bug with non-working custom ports
• Fixed a bug with replicas being incorrectly deleted from ZooKeeper for retained volumes

Note: Upgrade from the previous version will result in restart of ClickHouse clusters.

New features:

• Added 90s delay before restarting the stateful set, when ClickHouse server settings are modified. That makes sure ClickHouse server starts with an updated configuration. Before that there was a race condition between configmap updates and statefulset restart. The default timeout can be modified by 'spec.reconciling.configMapPropagationTimeout' property. See example
• Added 'troubleshooting' mode that allows pod to start even if ClickHouse server is failing. In troubleshooting mode liveness check is removed, and extra 'sleep' is added to the startup command. Controlled by 'spc.troubleshoot' property. See example
• Added a cleanup reconcilation logic that will remove k8s object that are labeled by a particular CHI but do not exist in the CHI manifest. The default behaviour can be altered with 'spec.reconciling.cleanup.unknownObjects' . See example
• ZooKeeper manifests have been modified to use 'standaloneEnabled=false' for single node setups as recommended in ZooKeeper documentation. ZooKeeper version has been bumped to 3.6.3

Bug fixes:

• Fixed a bug when replica has not been deleted in ZooKeeper when scaling down (#735). The bug has been introduced in 0.14.0
• Fixed a bug when PVCs modified outside of operator could be re-created during the reconcile (#730).
• Fixed a bug when schema was not created on newly added replica sometimes